About the job
About Flexion Robotics
Flexion Robotics develops the intelligence layer for next-generation humanoid robots. The company was founded by scientists with deep expertise in robot reinforcement learning from organizations such as Nvidia and ETH Zürich. Supported by international venture capital, Flexion Robotics has quickly moved from initial code to deploying real humanoid capabilities in just a few months.
Simulation Software Engineer Role
This Simulation Software Engineer position centers on building and maintaining the simulation stack that powers robot learning at Flexion Robotics in Zürich. Simulation is a core part of the company’s strategy for scaling intelligence. This role involves full ownership of the simulation platform, from physics fidelity to throughput optimization, with direct impact on how robots learn and perform.
Main Responsibilities
- Simulation stack ownership: Design, develop, and maintain the core simulation infrastructure for robot learning. Ensure its architecture, reliability, and evolution meet the team’s needs.
- Simulation fidelity: Build and improve models for contacts, actuators, deformables, and sensors. Focus on reducing the sim-to-real gap, increasing physical realism, and developing tools to validate simulations against real-world data.
- Training throughput optimization: Enhance the simulation stack for large-scale reinforcement learning. Implement efficient multi-GPU and multi-node pipelines, and identify and resolve bottlenecks to maximize training speed.
- Photorealistic rendering and synthetic data: Integrate and extend rendering pipelines to support vision-based learning. Enable domain randomization and generate synthetic datasets for robust perception systems.
- Software-in-the-loop testing: Maintain infrastructure for validating policies and systems within simulation before deployment. Support rapid iteration and reduce risks during hardware transitions.
- Collaboration: Work closely with engineers and researchers in reinforcement learning, control, perception, and hardware to ensure simulation aligns with real-world performance and deployment goals.

